Applied Clinical Trials Online recently published a new article from Ryan Bernhardt, General Manager of Virscidian at Dotmatics, making the case that lab orchestration has to mean more than coordinating instruments and robotics.
Bernhardt's argument: orchestration conversations usually stop at the physical layer, scheduling software moving samples through a queue. That framing misses the connective layer that actually holds a lab together. When experimental intent gets separated from results, scientists end up bridging the gap by hand, and that connective tissue is fragile.
A true definition of orchestration spans three domains at once:
Physical: instruments, robotics, and scheduling platforms that move samples and sequence tasks.
Logical: protocols, workflow design, and business rules, where scientific intent lives before it becomes action.
Digital: where results land alongside their context and stay connected across teams, time, and the inevitable turnover in tools and methods.
The cost of skipping this shows up differently depending on the role. Scientists lose a day or more per experiment manually pulling data together by hand. Digitalization leads inherit vendor integrations that were never built to talk to each other, patchwork that never finishes and scales badly. R&D directors watch technology investments underdeliver because each one lands on the same fragmented foundation.
Bernhardt places AI at the end of the sequence, not the beginning. Agentic AI is only as trustworthy as the data underneath it, and Gartner predicts 80% of agentic AI initiatives in life sciences will fail in 2026 due to gaps in transparency and evidence-based reasoning. In a related Dotmatics study, 80% of scientists said the workarounds required to turn CRO-generated data into usable output are hurting their work, and nearly 70% reported compromised decision-making as a result.
The piece closes with four questions R&D leaders can use to pressure-test their own environment today, including how much accumulated knowledge walks out the door when a scientist leaves, and whether an AI initiative could actually reason across the full scientific record, the intent behind experiments, the decisions made, and the context connecting them.
Read the full article for the complete argument and the reference architecture Bernhardt uses to distinguish a connected lab from an integration-dependent one.